Ceanothus Herbaceous) Growth and Reproduction

Am. Midl. Nat. 141:51–58

Effects of Fire, Browsers and Gallers on New Jersey Tea (Ceanothus herbaceous) Growth and Reproduction

HEATHER L. THROOP1 Department of Biology, Carleton College, Northﬁeld, Minnesota 55057

AND

PHILIP A. FAY Division of Biology, Ackert Hall, Kansas State University, Manhattan, Kansas 66506

ABSTRACT.—Woody plant species in grassland ecosystems can be subjected to damage from fire and multiple herbivore species, but interactions between fire and herbivory can modify their separate impacts on woody plant life histories. We studied how galling (by Periploca ceanothiella, Lepidoptera: Cosmopterigidae), deer browsing (Odocoilius virginianus) and fire affected the growth and reproduction of the woody shrub Ceanothus herbaceous (Rhamna- ceae) on a burned and an unburned site at Konza Prairie Research Natural Area in eastern Kansas. Fire was the major influence on C. herbaceous growth, causing plants to produce long unbranched vegetative ramets from protected belowground meristems, while unburned plants were heavily branched and bore shorter shoots and numerous inflorescences. Un- burned plants experienced higher gall frequencies, more galls on their longest shoots, but similar deer browsing compared to burned plants. Ramets with herbivore damage had more branches and inflorescences than undamaged ramets, especially where both herbivores were present. Ceanothus herbaceous’ flexible life history responses suggest tolerance of multiple forms of damage.

INTRODUCTION Fire and herbivory exert pervasive influences on the composition and productivity of many ecosystems, including the tallgrass prairies of North America. One major effect of fire is suppression of woody plants in favor of grasses and forbs (Bragg and Hulbert, 1976). Fire typically suppresses woody species by destroying aboveground biomass and meristems, which alters subsequent plant growth, architecture, reproductive capacity and competitive interactions (Glenn-Lewin et al., 1990; Yeaton and Bond, 1991; Matlack et al., 1993). In addition, fire modifies abiotic parameters important to plant growth, including temperature, light and resource availability (Old, 1969; Hulbert, 1988; Niesenbaum, 1992; Turner et al., 1997). Herbivory has many of the same influences on woody plants as fire including biomass removal, meristem death and reduced growth, reproduction and competitive ability (Fay and Hartnett, 1991; Erasmus et al., 1992; Brown, 1994). The severity of herbivore impacts on plants depends on the timing, intensity and type of damage inflicted, and a plant’s postdamage growth and physiological responses (Watson and Casper, 1984; Rosenthal and Kotanen, 1994; Fay et al., 1996). Herbivore impacts may be compounded when plants are damaged by several herbivore species (Willis et al., 1993). Fire can strongly affect the extent to which herbivore damage actually occurs. For ex- ample, fire affects which herbivore species are present (Warren et al., 1987). Some herbivore species tend to be killed in fires because their phenologies or life histories render them unable to escape. Other species mainly experience indirect effects such as postfire

1 Present address: Department of Ecology and Evolution, State University of New York, Stony Brook, 11794

51 52 THE AMERICAN MIDLAND NATURALIST 141(1) changes in host plant quality and abundance (Stein et al., 1992) or abiotic conditions (Hulbert, 1988). These direct and indirect effects of fires can affect herbivore abundances (Warren et al., 1987) and within-plant herbivore distributions (Rosenthal and Kotanen, 1994). As a result, the net impact on woody plant life histories of a fire/herbivore damage regime can be difficult to predict. New Jersey Tea (Ceanothus herbaceous: Rhamnaceae) is found primarily on prairies and open woodlands throughout the midwest and south, and locally in the eastern United States (Bartgis et al., 1997). It is an abundant shrub in Flint Hills tallgrass prairies (Gibson and Hulbert, 1987), where it persists despite regular damage from fire and herbivores. Fire damage typically occurs during spring prescribed burns, which are often conducted to re- duce woody species populations. In addition, two herbivores commonly damage C. herbaceous; a stem-galling moth (Periploca ceanothiella, Lepidoptera: Cosmopterigidae) whose abundance is directly influenced by spring fires (PAF and D.C. Hartnett, pers. observ.), and a browsing mammal, white-tailed deer (Odocoileus virginianus). The abundance of C. herbaceous despite consistent pressure from fire and multiple herbivores suggests that this species possesses effective tolerance mechanisms. To examine damage tolerance in C. herbaceous, we conducted field studies of plant growth characteristics in burned, galled and browsed plants in natural populations. We focused on these specific questions: (1) how does fire affect the incidence of galling and browsing, and within-plant gall distributions? (2) how do fire, galling and browsing affect plant growth form and reproduction?

METHODS Study site description.—Ceanothus herbaceous populations were studied at the Konza Prairie Research Natural Area, a 3487 ha tallgrass prairie in the northern part of the Kansas Flint Hills. Konza is divided into replicate watershed-sized experimental units assigned to factorial combinations of grazing (by cattle or bison) and burning (at 1, 2, 4, 10 or 20-yr intervals; see http://climate.konza.ksu.edu for detailed site map and descriptions). Konza’s topography is steeply dissected, with each watershed containing shallow-soiled uplands, steep rocky slopes and deep-soiled lowlands. The overall vegetation consists of a matrix of dominant warm-season grasses (primarily Andropogon gerardii, Sorghastrum nutans, Panicum virgatum and Schizachyrium scoparium), numerous forbs and several woody species (Freeman and Hulbert, 1985; Van Cleve and Martin, 1991). Plant community composition at a specific location depends strongly on its particular fire/grazing regime and topographic position (Gibson and Hulbert, 1987; Hartnett and Fay, 1998). Natural history.—Ceanothus herbaceous grows abundantly on watershed slopes at Konza (average cover ഡ 12% on LTER vegetation transects) regardless of fire frequency. Ceanothus herbaceous plants (i.e., genets) consist of multiple ramets originating from belowground meristems. During the first year of growth, each ramet is generally unbranched, usually 20– 25 cm tall, and bears numerous lateral meristems. In subsequent growing seasons, ramets become branched due to new shoots arising from the previous crop of lateral meristems. After several years, each ramet supports many shoots and can reach over 50-cm tall. Spring fires periodically reset plant growth form by killing all accumulated growth. This is followed by the rapid development of a new set of unbranched ramets. Ramets produce leaves in mid-to-late March, flower during April and May and set fruits by mid-June (Great Plains Flora Assoc., 1986). Periploca ceanothiella is an abundant stem galler on Ceanothus herbaceous. Beyond basic taxonomic information (Cosens, 1908; Hodges, 1978) the species is unstudied. Galls form in July and contain one larva which feeds and overwinters in the gall, emerging the following March. Galls persist on the plant after larval emergence (PAF and HLT pers. observ.), 1999 THROOP & FAY:NEW JERSEY TEA 53 allowing quantification of past galling history of plants. Morphologically, P. ceanothiella galls develop through a combination of inhibited internode elongation and lateral swelling. Galls are elliptical and retain the lateral buds from each node involved in the gall. Gall formation often prevents further terminal growth, and lateral shoots may arise from the gall’s lateral buds. White-tail deer browse Ceanothus herbaceous primarily during fall and winter, removing the terminal meristem from varying numbers of shoots per ramet. The amount of tissue lost to browsing cannot be determined after-the-fact, but the loss of the terminal meristem may alter C. herbaceous growth by increasing lateral branching at the expense of additional terminal growth. This is a common effect of herbivores in other woody plant species (Whi- tham and Mopper, 1985). Field studies.—We conducted field studies during June and July 1995 in two ungrazed watersheds with similar soil type and topography. One watershed has been burned every April since 1988 (Konza designation K1A, 100 ha), and the adjacent watershed (K20A, 90 ha) has been burned approximately once every 20 yr, and was last burned in 1991. These watersheds are typical of burned and unburned Konza watersheds in overall species composition, soils and presence of C. herbaceous and its herbivores. To assess Ceanothus herbaceous herbivory rates, growth characteristics and reproduction, we counted the numbers of herbivore damaged shoots, intact shoots and inflorescences on each ramet on plants in both watersheds. Plants were selected by marking four randomly- oriented transects per watershed, and then selecting the first ten plants per transect (n ϭ 40 plants per watershed, except n ϭ 25 plants for browsing data in K20A). For each plant we noted the numbers of inflorescences and undamaged, galled, or browsed shoots on each ramet. Herbivory was expressed as the percentage of ramets per plant galled or browsed. Data analysis.—We used a randomization procedure (Manly, 1991) to analyze burning effects on herbivory rates and plant growth and reproduction responses. This procedure was chosen because standard analysis of variance techniques are considered inappropriate when samples are not statistically independent (Hurlbert, 1984), as could be the case with one watershed per burn treatment. Briefly, the randomization procedure generates an expected frequency distribution of treatment sum of squares for the null hypothesis of no treatment effect by repeatedly (n ϭ 5000) assigning plants at random to two treatment groups and calculating sums of squares. The significance level of the test is the proportion of expected sums of squares that are less than the experimentally-observed sums of squares (Adams and Anthony, 1996). To measure herbivore impacts on plant growth and reproduction, we compiled numbers of shoots and inflorescences per ramet by herbivory type (browsed, galled or both), and analyzed these responses using one way analysis of variance. This analysis was conducted on the unburned watershed only, due to low numbers of galls and inflorescences on the annually burned watershed. To assess the effects of fire on the distribution of Periploca ceanothiella galls within Cea- nothus herbaceous, we compiled shoot length frequency distributions for galled and undamaged shoots in each watershed. These frequency distributions were based on intact shoot lengths, measured directly for undamaged shoots and estimated for galled shoots (Fay and Samenus, 1993). Intact lengths of galled shoots (i.e., their length had they not been galled) were estimated using bud count-shoot length regressions based on 200 randomly chosen shoots per watershed (for unburned plants length ϭ 0.95*buds ϩ 5.69, R2 ϭ 0.71, P Ͻ 0.001; for burned plants length ϭ 0.99*buds Ϫ 2.32, R2 ϭ 0.71, P Ͻ 0.001). For analysis, shoot lengths were grouped in 20 cm intervals, and the frequencies of estimated galled shoot lengths were compared to the total distributions with chi-square tests. 54 THE AMERICAN MIDLAND NATURALIST 141(1)

TABLE 1.—Shoot, ramet and inﬂorescence production and incidence of O. virginianus browsing and P. ceanothiella galls on annually burned and unburned C. herbaceous plants. All values are mean ϮSE. For all variables but browsing, 4999 of the randomly generated sums of squares were less than the observed, yielding P-values of 0.0002

Annually burned Unburned P value Ramets per plant 29.83 Ϯ 3.44 7.63 Ϯ 0.81 0.0002 Shoots per ramet 1.15 Ϯ 0.04 37.77 Ϯ 4.00 0.0002 Shoot length (cm) 32.84 Ϯ 0.34 16.96 Ϯ 0.41 0.0002 Inﬂorescences per plant 0.08 Ϯ 0.08 29.38 Ϯ 4.56 0.0002 % of ramets browsed 78.1 Ϯ 4.5 79.7 Ϯ 5.4 0.96 % of ramets galled 3.5 Ϯ 1.2 71.7 Ϯ 4.3 0.0002

RESULTS Influence of fire.—Annually-burned Ceanothus herbaceous plants produced many new ramets from belowground buds (Table 1). The new ramets were generally unbranched (i.e., composed of a single shoot), twice the length of the shoots produced on unburned ramets and bore no inflorescences. In contrast, unburned plants supported fewer ramets, but each ramet produced many shoots and inflorescences, resulting in a highly branched growth form. Influence of herbivory.—Galls rarely formed on ramets of burned plants but were found on nearly 75% of ramets in unburned plants (Table 1). In contrast, deer browsing occurred on nearly 80% of ramets in both burned and unburned plants. Galls were equally likely to occur on all shoot lengths in burned plants, but in unburned plants long shoots (Ͼ40 cm) were nearly twice as likely to be galled as shorter shoots (Table 2). Herbivory had the opposite effects of fire on Ceanothus herbaceous growth (Fig. 1). Dam- aged ramets produced more shoots and inflorescences than undamaged ramets. Ramets with damage from one herbivore showed increases in shoot and inflorescence numbers over undamaged ramets. Ramets damaged by both herbivores produced at least double the shoots and inflorescences of ramets damaged by a single herbivore.

TABLE 2.—Relationship between shoot length and galling rate for P. ceanothiella galls on burned and unburned C. herbaceous plants

Shoot length (cm) # shoots # galled (%) # expected ␹2 Annually burned plants Ͻ20 37 21 (56.7) 18.54 0.33 20–40 292 147 (50.3) 147.08 0.00 Ͼ40 80 38 (47.5) 40.38 0.14 Total 409 206 (50.4) 0.47 ns Unburned plants Ͻ20 291 141 (48.4) 146.27 0.19 20–40 86 39 (45.3) 43.24 0.42 Ͼ40 29 24 (82.7) 14.48 6.26 Total 406 204 (50.2) 6.87* *PϽ 0.05, df ϭ 2 1999 THROOP & FAY:NEW JERSEY TEA 55

FIG. 1.—Effects of P. ceanothiella galls and O. virginianus browsing on shoot and inflorescence production of unburned C. herbaceous ramets. All values are mean Ϯ SE.Bϭ browsed, G ϭ galled, U ϭ undamaged, B&G ϭ browsed and galled. Treatments with the same letters above columns are not significantly different. a) Shoots per ramet (F ϭ 35.7, P Ͻ 0.001). b) Inflorescences per ramet (F ϭ 12.14, P Ͻ 0.001)

DISCUSSION Fire markedly altered the growth of Ceanothus herbaceous. Immediately following a fire many long, vigorous, asexual ramets arose from protected basal buds. In contrast, the es- tablished ramets of unburned plants were highly branched, and new growth consisted of many shorter shoots and inflorescences. Vigorous asexual regrowth is typical of shrubs in fire-prone habitats (Stein et al., 1992; Romo et al., 1993), which often regenerate from protected belowground parts (Paige, 1992). Despite this vigorous postfire regrowth, cover of C. herbaceous is low immediately after fire (Gibson and Hulbert, 1987; Gibson, 1988) and increases in the years following fire as ramets become increasingly branched. By not producing inflorescences, annually burned Ceanothus herbaceous allocates resourc- es primarily to vegetative growth. This allocation pattern has short- and long-term benefits. Immediately after fire, vegetative regrowth would help C. herbaceous acquire space in the highly competitive tallgrass prairie plant canopy (Collins and Gibson, 1990). Where fires recur annually, strong vegetative growth may also help prevent local extinction of C. herbaceous populations because suppressed flowering could lead to low seedling recruitment due to few seeds in the seed bank (Whelan, 1995). Where fire is less frequent, seed production may be maximized in later growing seasons by early investment in vegetative growth, since there is usually a strong correlation between plant size and sexual reproductive effort (Samson and Werk, 1986; Hartnett, 1990). Fire reduced the frequency of Periploca ceanothiella galls but had no effect on deer browsing rates. Low gall frequencies on the burned watershed were most likely due to mortality of P. ceanothiella larvae in the spring fire. High mortality of larvae in galls during spring burns has been documented in other gallers of tallgrass prairie plants (Fay and Samenus, 1993). There was apparently little postfire immigration of P. ceanothiella adults into the burned watershed, even though potentially suitable shoots were available for galling and 56 THE AMERICAN MIDLAND NATURALIST 141(1) an abundant source of migrants was close by in the unburned watershed. We expected higher deer browsing rates on the burned watershed since postfire growth is often more nutritious and succulent than older plant tissue (Stein et al., 1992). However, since deer browse Ceanothus herbaceous primarily during fall and winter, plant quality differences between burned and unburned watersheds may have been diminished. Fire changed the within-plant distribution of shoot lengths and Periploca ceanothiella galls. Long shoots were less abundant and more frequently galled in the unburned watershed compared to the burned watershed. High gall frequencies on long shoots has been reported for other galling insects on woody plant species (Price et al., 1987). Fire thus appeared to affect the potential for galls to damage Ceanothus herbaceous in two ways. Unburned plants experienced both higher overall gall frequencies as well as disproportionate damage to long shoots, which have the greatest potential for future growth. Herbivore impacts on Ceanothus herbaceous were less severe than fire impacts. Whereas fire killed all aboveground tissues, both herbivores left live shoots and meristems from which C. herbaceous could continue to grow and flower. Galled and browsed ramets showed similar increases in branching. Even though their feeding styles differed, both herbivores caused damage that inhibited apical dominance, thus promoting similar lateral branching responses. However, browsing might be more harmful to C. herbaceous in the long term than galling because browsing removes part of C. herbaceous’ reserve meristem population and removes more biomass than galling. We summarize the net impacts of the fire/herbivore damage regime on Ceanothus herbaceous with a working hypothesis. Ceanothus herbaceous had a flexible, mainly fire-governed, life-history. Fire puts C. herbaceous in a vegetative unbranched state, and annual burning will maintain this state. With two or more years between fires, C. herbaceous becomes branched and allocation to sexual reproduction increases. Herbivory speeds the progression of branching and inflorescence production, but fire can mitigate herbivore impacts by re- ducing populations of one of C. herbaceous’ common herbivores. The flexible life history responses shown by Ceanothus herbaceous in response to fire and herbivores suggests tolerance of both forms of damage (Rosenthal and Kotanen, 1994), with fire tolerance conveyed by protected belowground meristems and herbivory tolerance conveyed by aboveground lateral meristems. Woody plants inhabiting fire-dominated ecosystems such as grasslands may commonly be subject to damage from fire and multiple herbivores. The response of C. herbaceous to fire and multiple herbivores suggests an important life history pattern by which woody plants can persist in such systems.

Acknowledgments.—HLT conducted this study while participating in the 1995 National Science Foun- dation Research Experience for Undergraduates program at Kansas State University (BIR 9322177). PAF thanks the Konza Prairie LTER program for additional support. DC Adams provided statistical advice and provided the SAS ﬁles to run the randomization program. Konza Prairie Research Natural Area is a preserve of the Nature Conservancy, managed for ecological research by the Division of Biology, Kansas State University.

LITERATURE CITED

ADAMS,D.C.AND C. D. ANTHONY. 1996. Using randomization techniques to analyse behavioural data. Anim. Behav. 51:733–738. BARTGIS, R L., G. P. FLEMING AND R. WIEGARD. 1997. The prairie-redroot (Ceanothus herbaceus RAF.) in the mid-Atlantic United States. Castanea, 62:127–128. BRAGG,T.B.AND L. C. HULBERT. 1976. Woody plant invasion of unburned Kansas bluestem prairie. J Range Manage., 29:19–23. 1999 THROOP & FAY:NEW JERSEY TEA 57

BROWN, D. G. 1994. Beetle folivory increases resource availability and alters plant invasion in monocultures of goldenrod. Ecology, 75:1673–1683. COLLINS,S.L.AND D. J. GIBSON. 1990. Effects of fire on community structure in tallgrass and mixed- grass prairie. p. 81–98. In: S. L. Collins and L. L. Wallace (eds.). Fire in North American tallgrass prairies. University of Oklahoma Press, Norman. COSENS, A. 1908. A new lepidopterous gall-producer. Can. Entomol., 40:107–108. ERASMUS, D. J., P. H. BENNETT AND J. van STADEN. 1992. The effect of galls induced by the gall fly Procecidochares utilis on vegetative growth and reproductive potential of crofton weed, Ageratina adenophora. Ann. Appl. Biol., 120:173–181. FAY,P.A.AND D. C. HARTNETT. 1991. Constraints on growth and allocation patterns of Silphium integrifolium (Asteraceae) caused by a cynipid gall wasp. Oecologia, 88:243–250. ,D.C.HARTNETT AND A. K. KNAPP. 1996. Plant tolerance of gall-insect attack and gall-insect performance. Ecology, 72:521–534. , AND R. J. SAMENUS, Jr. 1993. Gall wasp (Hymenoptera: Cynipidae) mortality in a spring tallgrass prairie fire. Environ. Entomol., 22:1333–1337. FREEMAN,C.C.AND L. C. HULBERT. 1985. An annotated list of the vascular flora of the Konza Prairie Research Natural Area, Kansas. Trans. Kans. Acad. Sci., 88:84–115. GIBSON, D. J. 1988. Regeneration and fluctuation of tallgrass prairie vegetation in response to burning frequency. Bull. Torr. Bot. Club, 115:1–12. AND L. C. HULBERT. 1987. Effects of fire, topography and year-to-year climatic variation on species composition in tallgrass prairie. Vegetatio, 72:175–185. GREAT PLAINS FLORA ASSOCIATION. 1986. Flora of the Great Plains. University Press of Kansas, Lawrence, 1392 p. GLENN-LEWIN, D. C., L. A. JOHNSON,T.W.JURIK,A.AKEY,M.LEOSHKE AND T. ROSBURG. 1990. Fire in central North American grasslands: vegetative reproduction, seed germination, and seedling establishment. p. 28–45 In: S. L. Collins and L. L Wallace (eds.). Fire in North American tallgrass prairies. University of Oklahoma Press, Norman. HARTNETT, D. C. 1990. Size-dependent allocation to sexual and vegetative reproduction in four clonal composites. Oecologia, 84:254–259. AND P. A. FAY. 1998. Plant populations: patterns and processes. p. 81–99 In: A. K. Knapp, J. M. Briggs, D. C. Hartnett and S. L. Collins (eds.). Grassland dynamics: long-term ecological research in tallgrass prairie. Oxford University Press, New York. HODGES, R. W. 1978. Fasc. 6.1, Gelechioidea (in part): Cosmopterigidae. In: R. B. Dominick, T. Dominick, D. C. Ferguson, J. G. Franclemont, R. W. Hodges and E. G. Munroe (eds.) The moths of America north of Mexico. E.W. Classey, Ltd., London. 166 p. HULBERT, L. C. 1988. Causes of fire effects in tallgrass prairie. Ecology, 69:46–58. HURLBERT, S. H. 1984. Pseudoreplication and the design of ecological field experiments. Ecol. Monogr., 54:187–211. MANLY, B. F. J. 1991. Randomization and Monte Carlo methods in biology. Chapman and Hall, New York. 281 p. MATLACK, G. R., D. J. GIBSON AND R. E. GOOD. 1993. Regeneration of the shrub Gaylussacia baccata and associated species after low-intensity fire in an Atlantic coastal plain forest. Am. J. Bot., 80: 119–126. NIESENBAUM, R. A. 1992. The effects of light environment on herbivory and growth in the dioecious shrub Lindera benzoin (Lauraceae). Am. Midl. Nat., 128:270–275. OLD, S. M. 1969. Microclimate, fire, and plant production in an Illinois prairie. Ecol. Monogr., 39:355– 384. PAIGE, K. N. 1992. The effects of fire on scarlet gilia: an alternative selection pressure to herbivory? Oecologia, 92:229–235. PRICE, P. W., H. ROININEN AND J. TAHVANAINEN. 1987. Why does the bud-galling sawfly, Euura mucronata, attack long shoots? Oecologia, 74:1–6. ROMO, J. T., P. L. GRIZL,R.E.REMAN AND E. A. DRIVER. 1993. Standing crop, biomass allocation patterns 58 THE AMERICAN MIDLAND NATURALIST 141(1)

and soil-plant water relations in Symphoricarpos occidentalis Hook. following autumn or spring burning. Am. Midl. Nat., 130:106–115. ROSENTHAL,J.P.AND P. M. KOTANEN. 1994. Terrestrial plant tolerance to herbivory. Trends Ecol. Evol., 9:145–148. SAMSON,D.A.AND K. S. WERK. 1986. Size-dependent effects in the analysis of reproductive efforts in plants. Am. Nat., 127:667–680. STEIN, S. J., P. W. PRICE,W.G.ABRAHAMSON AND C. F. SACCHI. 1992. The effect of fire on stimulating willow regrowth and subsequent attack by grasshoppers and elk. Oikos, 65:190–196. TURNER, C. L., J. M. BLAIR,R.J.SCHARTZ AND J. C. NEEL. 1997. Soil N and plant responses to fire, topography, and supplemental N in tallgrass prairies. Ecology, 78:1832–1843. VAN CLEVE,K.AND S. MARTIN. 1991. Long-term ecological research in the United States: a network of research sites, 6th ed. revised. LTER Network Office, Seattle. 178 p. WARREN, S. D., C. J. SCIFRES AND P. D. TEEL. 1987. Response of grassland arthropods to burning: a review. Agric. Ecosys. Environ., 19:105–130. WATSON,M.A.AND B. B. CASPER. 1984. Morphogenetic constraints on pattern of carbon distributions in plants. Annu. Rev. Ecol. Sys., 15:233–258. WHELAN, R. J. 1995. The ecology of fire. Cambridge University Press, Cambridge. 346 p. WHITHAM,T.G.AND S. MOPPER. 1985. Chronic herbivory: impacts on architecture and sex expression of pinyon pine. Science, 228:1089–1091. WILLIS, A. J., J. E. ASH AND R. H. GROVES. 1993. Combined effects of two arthropod herbivores and water stress on growth of Hypericum species. Oecologia, 96:517–525. YEATON,R.I.AND W. J. BOND. 1991. Competitive interactions between two shrub species: dispersal differences and fire promote coexistence. Am. Nat., 138:328–341.

SUBMITTED 29 SEPTEMBER 1997 ACCEPTED 30 APRIL 1998